Key takeaways
- Static testing examines work products without running them: requirements, designs, code, test cases. Dynamic testing runs the software.
- It finds defects directly, often before they can cause any failure, which makes them very cheap to fix.
- It's done by people (reviews) and by tools (static analysis). Both are part of a tester's job.
Static vs dynamic
Most of the testing you've seen so far is dynamic: you run the software and watch what happens.
Static testing looks at things without running them. You read a requirement and spot a contradiction. A tool scans code and flags a variable that's never used. Nothing is executed, yet a defect is found.
| Static testing | Dynamic testing | |
|---|---|---|
| Runs the software? | No | Yes |
| Finds | Defects directly | Failures, then you look for the defect |
| Works on | Any work product: requirements, designs, code, tests | Running software only |
| When | From the very first document | Once something can run |
What you can review
Almost anything a team produces:
- Requirements and user stories: are they clear, complete, consistent and testable?
- Designs and architecture: are risky areas, like integrations and data flows, thought through?
- Code: is it correct, readable and secure?
- Test cases and test plans: do they cover the risks? Are expected results right?
- Contracts, help texts, configuration files.
Types of review
The ISTQB syllabus describes several types, from informal to formal:
| Type | What it looks like | Good for |
|---|---|---|
| Informal review | A colleague takes a quick look; no fixed process | Everyday feedback, small changes |
| Walkthrough | The author leads the group through the document | Sharing understanding, gathering ideas |
| Technical review | Technical peers check the work against standards | Design and code decisions |
| Inspection | Most formal: defined roles, checklists, metrics | Safety-critical or high-risk work |
In most teams today, the most common review is the pull request review: before code is merged, at least one other person reads it.
Static analysis: reviews by tools
Tools read code and flag problems automatically:
- Compilers and linters: syntax errors, unused variables, unreachable code, suspicious patterns.
- Type checkers: values used as the wrong type.
- Security scanners (SAST): insecure patterns, like building database queries from user input.
They're fast, tireless and consistent. They run on every change in the pipeline. But they can't judge whether the code does what the business needs. That still takes people.
In February 2014, Apple released urgent updates for iOS and macOS. The bug, nicknamed "goto fail", was in the code that checks secure (SSL/TLS) connections.
In one function, the line goto fail; appeared twice in a row. The second copy always ran, so the program jumped past a key verification step. Connections that should have been rejected could be accepted, allowing an attacker on the same network to intercept supposedly secure traffic.
The defect is visible to a careful reader, and tools that warn about unreachable code can flag the skipped check.
Lesson: some serious defects are easier to find by reading than by running. A code review or a static analysis warning, taken seriously, is cheap compared with an emergency security update.
How to review well
- Know the goal. Are you looking for ambiguity, missing cases, security issues? A focus finds more than "have a look".
- Use a short checklist: unclear words ("fast", "user-friendly"), missing error cases, numbers without units, boundaries not defined.
- Review in small pieces. An hour on a few pages finds more than a day skimming everything.
- Comment on the work, not the person. "This step doesn't say what happens if payment fails", not "you forgot".
- Follow up. A review that changes nothing was wasted time.
Do
Ask to review user stories before the sprint starts. One question at that stage can save days of rework later.
Don't
Treat reviews as a formality and approve without reading. A rubber-stamped review gives false confidence.
Check yourself
A requirement says: 'The report should load quickly and show all relevant data.' What would you comment in a review?
Show answer
Both key words are untestable: "quickly" needs a number (for example, "within 3 seconds for up to 10,000 rows"), and "all relevant data" needs a list of fields. Ask the author to make them specific. Finding this in a review takes minutes; finding it after development can mean rework.
